Sep 22, 2026 · 6 min listen · Last updated September 22, 2026
From storyflo. This is your daily audio brief. Theo, September 22nd. The systems update — five tech stories that bear on what's coming next. Let's get into it. First, from TechCrunch. Morphotonics raises €40M to expand its display tech into data centers.
Listen · storyflo · tech
Daily Tech Brief · September 22nd
0:00-6:17
Pick your daily storyteller
Subscribe to match with Theo, Jessica, Chloe, Mason, Brock — your voice, every brief.
Audio pre-rendered by Storyflo · cached + delivered from the edge
Morphotonics raises €40M to expand its display tech into data centers
The Netherlands’ ASML is possibly the biggest name in lithography, using its machines to etch patterns onto silicon wafers inside a computer chip factory, known as a “fab.” Fellow Dutch company Morphotonics is smaller and works in a different corner of the same field: nanoimprint lithography, or NIL, a process used to manufacture the displays in AR glasses and other optics. It said it has benefited from ASML’s supply chain and uses many of the same vendors. Morphotonics has spent 12 years developing that technology, largely under the radar.
Here in 2026 if we need a terminal the chances are we fire one up on our computer and continue on our way without a thought. What we are doing of course is emulating the experience of earlier hardware terminals, ones with a serial port on the back, a keyboard, a screen, or if they are a bit older, a teletype printer. [Sheila Dixon] is recreating that experience for the RC2014 retrocomputer, and she’s using some period-appropriate hardware and techniques. Perhaps the most obvious thing about her build is her choice of printer.
I Built a Benchmark for My Agent. The Smaller Model Won.
Every AI application that wraps an agent is a harness! In LangChain’s Terminal-Bench experiment, changing only the harness (with the same model) moved a coding agent from ~30th place into the top 5: the harness, not the model, is what makes a coding agent good. In the open-source course Building a Coding Agent From Scratch, you’ll build that harness from scratch in Python: Decode, a complete coding agent that grows lesson by lesson from a bare agent loop into a swarm of remote agents running in parallel in the cloud.
Meet the Tech Bros and Gals — A retro animated cartoon series poking fun at real-life, well-known Silicon Valley entrepreneurs and venture capitalist billionaires
Animated cartoon mood and character brief Under consideration: Tech Bros, The Daily Cryptonite (Daily Planet wink from the Clark Kent & Lois Lane years), or another name — still open. Under consideration: Tech Bros, The Daily Cryptonite (Daily Planet wink from the Clark Kent & Lois Lane years), or another name — still open.
A new AI model called Jev has emerged from one of the co-founders of ChatGPT, and it’s creating quite a buzz. What’s striking about Jev is its focus—it doesn’t aim to compete with existing models like ChatGPT or Claude. Instead, it’s designed to handle the smaller, often overlooked decisions that arise in AI workflows. While models like Claude generate content, Jev is all about making choices: determining urgency, selecting tools, and deciding when a task is complete.
Think of Jev as a second brain that complements your main AI. It’s built to streamline processes by managing the nitty-gritty decisions that can bog down more complex models. This could be a game-changer for efficiency, especially in tasks that require quick, straightforward answers rather than extensive content creation.
The economics behind Jev are also noteworthy—at just $0.042 per million tokens, it’s incredibly cost-effective. The guide accompanying Jev breaks down everything from its decision-making modes to practical setup tips, making it accessible for anyone looking to enhance their AI capabilities. It’s an intriguing shift in how we think about AI’s role in our workflows, and I can’t help but wonder how it might change the way we interact with technology moving forward.
Meta's New 'Personal AI Agent' Muse Beats ChatGPT in Downloads - and Get Blocked by Amazon
Meta's new personal AI agent, Muse, has quickly climbed to the top of the iOS free app charts in the U.S., surpassing ChatGPT with over 2.5 million downloads in just under two weeks. The app is designed to help users manage various online tasks, like organizing emails and filling out forms, aiming to make AI more accessible for everyday people. This surge in popularity has even boosted Meta's stock by over 11%.
However, Muse is facing pushback from Amazon, which has blocked the app from accessing its store. Amazon claims that Muse didn't disclose its intent to browse the site and raises concerns about privacy and security, as it appears to capture customer credentials. This isn't Amazon's first encounter with AI agents; they've been actively trying to keep them off their platform, having previously sued other companies for similar reasons.
Interestingly, while Amazon is in talks with Meta about the issue, Shopify has chosen to partner with Muse, allowing it to facilitate checkouts on its platform. This partnership highlights the contrasting approaches different companies are taking toward AI integration.
Jev introduces a new shape of LLM - System One, aka Decision Models
Jev, the latest from TypeSafe AI, introduces a fresh approach to large language models, dubbed “System One” or decision models. Unlike traditional LLMs that generate text, Jev outputs floating point numbers that represent probabilities for yes/no questions, choices, or scoring systems. This means you can input various data forms and receive confidence scores that help classify or rank information. It’s designed to be efficient and cost-effective, charging only for input tokens, making it cheaper than many competitors.
The model can handle multiple questions simultaneously, which is a neat twist for users looking to streamline their queries. However, this shift towards a black box model raises concerns about transparency and bias, as the outputs don’t provide insight into the reasoning behind the scores. This could be problematic in sensitive applications like hiring, where hidden biases could skew results without clear explanations.
Despite these concerns, the community is buzzing with creative applications for Jev, from chat models to games. The excitement around its potential uses is palpable, and it’s fascinating to see how quickly people are experimenting with this new tool. As we navigate this evolving landscape, it’s clear that rigorous testing and evaluation will be crucial to ensure responsible deployment.
Apple challenges OpenAI’s forensic analysis in trade secrets case, seeks access to hardware R&D
Apple has asked the court overseeing its lawsuit against OpenAI and other defendants to let its own experts review forensic images used in the case, while also seeking documents related to parts of OpenAI’s hardware R&D. Here are the details.
Disturbing Experiment Points to Dangers of Using AI Models Not Meant for Robotics
In a creepy video, a robot arm powered by OpenAI’s GPT-6 Astra AI model picks up a large knife and pokes a baby doll sitting near a baguette, following instructions to “stab the thing that’s not the bread.” In another, a robot arm, this time using Anthropic’s Claude Fable 5.1 as its brain, carries out the frightening command to “put the screwdriver in the toaster.” The experiments were part of a safety benchmark created by the independent evaluation firm Robocurve.